Hidden Canvas: Synthesis
Details
Download Files
Model description
⚠️ Before You Generate — Please Read
Synthesis is not a plug-and-play model.
It rewards intention, patience, and understanding of how it works.
Before generating, please take a moment to read the full description and technical notes. It contains essential guidance on artifacts, LoRA usage, and how to get the best results.
Respect the craft — and the model will respect your vision.
Hidden Canvas: Synthesis
A second genesis — not a sequel, but a new step.
Built upon the foundation of Hidden Canvas, this model is not just an update. It is the awakening of everything that once lay dormant — refined, sharpened, reborn.
This model is both echo and evolution. It inherits the soul of the original, yet speaks in its own voice. It calls you to go deeper — beyond form, beyond prompt, beyond habit. What once whispered now speaks clearly. But the price remains the same: time, precision, and a willingness to fail before you flourish.
Hidden Canvas: Synthesis does not replace the first. It stands beside it — calm and powerful. Not a shadow, but a second flame. Where the first tested your patience, this one rewards your growth.
You’ve walked this path before.
Now walk it again — differently.
And this time, further.
Annotation:
Think you've mastered Hidden Canvas?
Synthesis will challenge everything you thought you knew. It doesn’t just ask for skill — it asks for evolution.
This model comes alive in the hands of those who follow instinct, break their own rules, and risk imperfection for something greater. If you want to create not just striking — but living images, Synthesis will meet you there.
But only if you're ready to let go of the formula.
Annotation (Technical Note):
Occasionally, you may notice a faint square artifact when using ADetailer — especially in portraits or darker compositions.
This is a known side effect, not a flaw in the model itself, but a byproduct of how face correction is applied. Think of it as a necessary sacrifice — one that can be minimized.
To reduce its visibility, adjust your Inpaint denoising strength to around 0.2–0.3.
This small tweak helps blend the correction more naturally into the base image.
Perfection takes effort.
Even shadows must be tamed.
Also, use LoRAs with great caution.
Synthesis was designed to be self-sufficient — balanced, expressive, and refined out of the box.
Adding external LoRAs without careful control of their strength can distort the model’s character and lead to poor results.
I don’t believe in patchwork fixes.
I believe a model should be beautiful on its own — without crutches or interference.
This isn’t a restriction. It’s an invitation: try it first as it is.
For the face, it is necessary to use ADetailer
Watch video example - https://civitai.com/images/70459219
For full-length works or long-range plans. The examples use Img2img - Inpaint
Samplers:
Recommended:
Sampler: Euler a / DPM++ SDE + Karras
Sampling Steps: 25-45
CFG Scale: 5-7 (For portraits: 5-6)